1. PCL (Point Cloud Library)

主要功能：支持2D/3D图像和点云处理
PCL Android资源
主要功能模块
PCL教材
PCL依赖库

库名	最小版本	描述
Eigen	3.0	高层次的C ++库，有效支持线性代数，矩阵和矢量运算，数值分析及其相关的算法，功能列表： Core：Matrix和Array类，基础的线性代数运算和数组操作 Geometry：旋转、平移、缩放、2D/3D旋转(Quanternion, AngleAxis) LU：求逆、行列式和LU分解 Cholesky：LLT和LDLT Cholesky分解 SVD：SVD分解（JacobiSVD） QR：QR分解 Eigenvalues：求特征值和特征向量
FLANN	1.7.1	Fast Library for Approximate Nearest Neighbors (近似近邻快速库) 一个在高维空间中执行快速近似近邻搜索的库最适合最近邻搜索的算法，以及一个根据数据集自动选择最佳算法和最佳参数的系统
Boost	1.40 (without OpenNI) 1.47 (with OpenNI)	为C++语言标准库提供扩展的一些C++程序库的总称 Boost库是一个可移植、提供源代码的C++库，作为标准库的后备，是C++标准化进程的开发引擎之一现有的 Boost 包含大约150种不同的函数库
VTK	5.6	PCL使用的可视化工具 (Visualization Toolkit) 支持3D渲染、3D交互

PCL编译

# flann库
sudo apt-get install libflann1.8 libflann-dev 
# eigen库
sudo apt-get install libeigen3-dev
# boost库  
sudo apt-get install libboost-all-dev 
# VTK 
sudo apt-get install libvtk5.10-qt4 libvtk5.10 libvtk5-dev

# 下载PCL
git clone https://github.com/PointCloudLibrary/pcl.git
#编译
cd pcl
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=/usr -DBUILD_apps=ON ..
make -j4
# install
sudo make install

# test
pcl_viewer model.pcd

1.1 基于pcl::visualization::PCLVisualizer 实时显示点云流的代码

include <pcl/point_types.h>
#include <pcl/visualization/cloud_viewer.h>



boost::mutex updateModelMutex;

boost::shared_ptr<pcl::visualization::PCLVisualizer> simpleVis (pcl::PointCloud<pcl::PointXYZ>::ConstPtr cloud)
{
        // --------------------------------------------
        // -----Open 3D viewer and add point cloud-----
        // --------------------------------------------
        boost::shared_ptr<pcl::visualization::PCLVisualizer> viewer (new pcl::visualization::PCLVisualizer ("3D Viewer"));
        viewer->setCameraPosition(0, 0, 0, 0, 0, 0, 0, 0, -1);
        viewer->setBackgroundColor(0,0,0);
        viewer->addPointCloud<pcl::PointXYZ> (cloud, "sample cloud");
        viewer->setPointCloudRenderingProperties (pcl::visualization::PCL_VISUALIZER_POINT_SIZE, 1, "sample cloud");
        viewer->initCameraParameters ();
        return (viewer);
}

void viewerRunner(boost::shared_ptr<pcl::visualization::PCLVisualizer> viewer)
{
        while (!viewer->wasStopped ())
        {
               viewer->spinOnce (100);
               boost::this_thread::sleep (boost::posix_time::microseconds (100));
        }
}

void main()
{      
	    pcl::PointCloud<pcl::PointXYZ>::Ptr cloud_ptr (new pcl::PointCloud<pcl::PointXYZ>);
        pcl::PointCloud<pcl::PointXYZ> &pcloud1 = *cloud_ptr;
        boost::shared_ptr<pcl::visualization::PCLVisualizer> viewer (new pcl::visualization::PCLVisualizer ("3D Viewer"));
        
        pcloud1.points.push_back( pcl::PointXYZ(10, 10, 80) );
        pcloud1.width = cloud_ptr->size();
        pcloud1.height = 1;
        pcloud1.is_dense = true;
        viewer = simpleVis(cloud_ptr);

        boost::thread vthread(&viewerRunner,viewer);
       
        while(1)//循环抓取深度数据
        {
              pcloud1.clear();
              for ( int _row = 0; _row < disp.rows; _row++ )
              {
                   for ( int _col = 0; _col < disp.cols; _col++ )
                   {
                       float x, y, z;
                       pcl::PointXYZ ptemp(x, y, z);
                       pcloud1.points.push_back( ptemp );
                    }
               }
               pcloud1.width = cloud_ptr->size();
               pcloud1.height = 1;
               pcloud1.is_dense = true;
               boost::mutex::scoped_lock updateLock(updateModelMutex);
               viewer->updatePointCloud<pcl::PointXYZ>(cloud_ptr,"sample cloud");
               updateLock.unlock();
               boost::this_thread::sleep (boost::posix_time::microseconds (100));
       }
       vthread.joint();
}

int main()
{
	pcl::visualization::PCLVisualizer viewer(“Cloud”);
	pcl::PointCloud::Ptr cloud(new pcl::PointCloud);
	int frame=0;
	while(1)
	{
		char buf[233];
		if (frame != 0)
		{
			sprintf(buf, “frame%d”, frame);
			viewer.removePointCloud(buf);
		}
		sprintf(buf, “frame%d”, ++frame);
		viewer.addPointCloud(cloud, buf);
		viewer.spinOnce();
	}
	return 0;
}

2. Open3D

Open3D是Intel继OpenCV又一开源巨作
官方文档
主要特点
- 天生支持最爱的Python
- 采用Google Filament作为渲染引擎，可在Android, Linux, MacOS和Windows上运行
- 聚焦可视化和渲染

2.1 编译环境

Ubuntu 18.04+: GCC 5+, Clang 7+
macOS 10.14+: XCode 8.0+
Windows 10 (64-bit): Visual Studio 2019+

2.2 编译过程 (Ubuntu 18.04)

下载代码

git clone --recursive https://github.com/intel-isl/Open3D

安装依赖

sudo apt install nasm  
util/scripts/install-deps-ubuntu.sh

2.3 设置Python绑定环境 (pybind11)

设置正确的Python执行文件

# Ubuntu 18.04默认安装了 Python 3.6.9
# 检查python版本
which python3
python3 -V
# 在CMake中增加
-DPYTHON_EXECUTABLE=/usr/bin/python3

支持Jupyter可视化

# 安装npm
sudo apt install nodejs
sudo apt install npm
#check npm
node -v
npm -v

2.4 编译过程

mkdir build; cd build

Ubuntu 和 MacOS

cmake -DCMAKE_INSTALL_PREFIX=/usr/local/open3d -DGLIBCXX_USE_CXX11_ABI=OFF -DPYTHON_EXECUTABLE=/usr/bin/python3 ..

Windows

cmake -DCMAKE_INSTALL_PREFIX=/usr/local/open3d \ 
      -DGLIBCXX_USE_CXX11_ABI=OFF \
      -DPYTHON_EXECUTABLE=/usr/bin/python3 \
	  -DBUILD_EIGEN3=ON  \
      -DBUILD_FLANN=ON   \
      -DBUILD_GLEW=ON    \
      -DBUILD_GLFW=ON    \
      -DBUILD_PNG=ON     \
      ..

执行编译

make -j4

安装和卸载C++库

# 安装
sudo make install
# 卸载
sudo make uninstall

安装和卸载Python包

# Create Python package in build/lib/python_package
make python-package
# Create pip wheel in build/lib/python_package/pip_package
make pip-package
# Create conda package in build/lib/python_package/conda_package
make conda-package
# Install pip wheel
sudo make install-pip-package

测试导入Open3D

python -c "import open3d"

2.5 阻塞和非阻塞示例代码

# -*-coding: utf-8 -*-
import open3d
import numpy as np
import cv2

# 绘制open3d座标系
axis_pcd = open3d.create_mesh_coordinate_frame(size=0.5, origin=[0, 0, 0])
# 在3D座标上绘制点：座标点[x,y,z]对应R，G，B颜色
points = np.array([[0.1, 0.1, 0.1], [1, 0, 0], [0, 1, 0], [0, 0, 1]])
colors = [[1, 1, 1], [1, 0, 0], [0, 1, 0], [0, 0, 1]]
 
test_pcd = open3d.geometry.PointCloud()  # 定义点云
 
# 方法1（非阻塞显示）
vis = open3d.Visualizer()
vis.create_window(window_name="Open3D1")
vis.get_render_option().point_size = 3
first_loop = True
# 先把点云对象添加给Visualizer
vis.add_geometry(axis_pcd)
vis.add_geometry(test_pcd)
while True:
    # 给点云添加显示的数据
    points -= 0.001
    test_pcd.points = open3d.utility.Vector3dVector(points)  # 定义点云座标位置
    test_pcd.colors = open3d.Vector3dVector(colors)  # 定义点云的颜色
    # update_renderer显示当前的数据
    vis.update_geometry()
    vis.poll_events()
    vis.update_renderer()
    cv2.waitKey(100)
 
# 方法2（阻塞显示）：调用draw_geometries直接把需要显示点云数据
test_pcd.points = open3d.utility.Vector3dVector(points)  # 定义点云座标位置
test_pcd.colors = open3d.Vector3dVector(colors)  # 定义点云的颜色
open3d.visualization.draw_geometries([test_pcd] + [axis_pcd], window_name="Open3D2")

2.6 添加自定义的 draw_geometries 函数

import os
import numpy as np
from open3d import *
points = np.random.rand(10000, 3)
point_cloud = PointCloud()
point_cloud.points = Vector3dVector(points)
draw_geometries([point_cloud])

# or
def custom_draw_geometry_with_key_callback(pcd):
    def change_background_to_black(vis):
        opt = vis.get_render_option()
        opt.background_color = np.asarray([0, 0, 0])
        return False
    def load_render_option(vis):
        vis.get_render_option().load_from_json(
                "../../TestData/renderoption.json")
        return False
    def capture_depth(vis):
        depth = vis.capture_depth_float_buffer()
        plt.imshow(np.asarray(depth))
        plt.show()
        return False
    def capture_image(vis):
        image = vis.capture_screen_float_buffer()
        plt.imshow(np.asarray(image))
        plt.show()
        return False
    key_to_callback = {}
    key_to_callback[ord("K")] = change_background_to_black
    key_to_callback[ord("R")] = load_render_option
    key_to_callback[ord(",")] = capture_depth
    key_to_callback[ord(".")] = capture_image
    draw_geometries_with_key_callbacks([pcd], key_to_callback)

custom_draw_geometry_with_key_callback(point_cloud)

2.7 体素下采样 (Voxel downsampling)

print("Downsample the point cloud with a voxel of 0.05")
downpcd = pcd.voxel_down_sample(voxel_size=0.05)
o3d.visualization.draw_geometries([downpcd], zoom=0.3412,
                                  front=[0.4257, -0.2125, -0.8795],
                                  lookat=[2.6172, 2.0475, 1.532],
                                  up=[-0.0694, -0.9768, 0.2024])

2.8 平面分割 (Plane segmentation)

使用RANSAC进行平面分割

pcd = o3d.io.read_point_cloud("../../TestData/fragment.pcd")
plane_model, inliers = pcd.segment_plane(distance_threshold=0.01,
                                         ransac_n=3,
                                         num_iterations=1000)
[a, b, c, d] = plane_model
print(f"Plane equation: {a:.2f}x + {b:.2f}y + {c:.2f}z + {d:.2f} = 0")

inlier_cloud = pcd.select_by_index(inliers)
inlier_cloud.paint_uniform_color([1.0, 0, 0])
outlier_cloud = pcd.select_by_index(inliers, invert=True)
o3d.visualization.draw_geometries([inlier_cloud, outlier_cloud], zoom=0.8,
                                  front=[-0.4999, -0.1659, -0.8499],
                                  lookat=[2.1813, 2.0619, 2.0999],
                                  up=[0.1204, -0.9852, 0.1215])

2.9 删除隐藏点

print("Convert mesh to a point cloud and estimate dimensions")
pcd = o3dtut.get_armadillo_mesh().sample_points_poisson_disk(5000)
diameter = np.linalg.norm(np.asarray(pcd.get_max_bound()) - np.asarray(pcd.get_min_bound()))
o3d.visualization.draw_geometries([pcd])

print("Define parameters used for hidden_point_removal")
camera = [0, 0, diameter]
radius = diameter * 100

print("Get all points that are visible from given view point")
_, pt_map = pcd.hidden_point_removal(camera, radius)

print("Visualize result")
pcd = pcd.select_by_index(pt_map)
o3d.visualization.draw_geometries([pcd])

3. 安装cmake 3.12.2或更高版本

# 删除已经安装的CMake旧版本（非必需项）：
apt-get autoremove cmake1
#下载cmake-3.12.2-Linux-x86_64.tar.gz安装包：
wget https://cmake.org/files/v3.12/cmake-3.12.2-Linux-x86_64.tar.gz
# 解压请运行以下命令：
tar zxvf cmake-3.12.2-Linux-x86_64.tar.gz
# 创建软链接
#   注: 文件路径是可以指定的，一般选择在/opt或者是/usr路径下，这里选择/opt：
mv cmake-3.12.2-Linux-x86_64 /opt/cmake-3.12.2
ln -sf /opt/cmake-3.12.2/bin/*  /usr/bin/
# 然后执行命令检查一下：
cmake --version

# cmake gui
cmake-gui

深度图+点云常用工具汇总

1. PCL (Point Cloud Library)

1.1 基于pcl::visualization::PCLVisualizer 实时显示点云流的代码

2. Open3D

2.1 编译环境

2.2 编译过程 (Ubuntu 18.04)

2.3 设置Python绑定环境 (pybind11)

2.4 编译过程

2.5 阻塞和非阻塞示例代码

2.6 添加自定义的 draw_geometries 函数

2.7 体素下采样 (Voxel downsampling)

2.8 平面分割 (Plane segmentation)

2.9 删除隐藏点

3. 安装cmake 3.12.2或更高版本

計算機視覺-數據集

ToF基礎知識

決策樹和隨機森林（Decision Trees and Random Forests）

OpenCV4 常用技巧

UDT Java 封裝

https://yachay.unat.edu.pe/blog/index.php?comment_area=format_blog&comment_component=blog&comment_co

linux以太網驅動總結